Prev: HELP: Somers' D
Next: v-fold cross validation in Clustering
From: Anders M�rup Jensen on 23 Mar 2010 05:44
I am trying to make a system where it would be convenient to convert a
Windows\DOS filename expression with wild cards to an equivalent Perl
regular expression.

Examples of DOS filenames:

*.*
?er*.sas
r?p.*s
e??d*.r*
??.?

Does anybody happen to have, or know how to create, a SAS function (or a SAS
macro) that can do the conversion ?

Anders M�rup





From: Patrick on 23 Mar 2010 07:57
If you're using SAS9.2 then look up PROC FCPM - else a function style
macro could do.

Not sure what you exactly want to do - but I assume you would mainly
need some logic in this function to convert "wild card" characters to
appropriate RegEx "characters".
From: Anders M�rup Jensen on 25 Mar 2010 03:40

"Anders M�rup Jensen" <amo(a)ssi.dk> skrev i en meddelelse
news:newscache$r2bqzk$90a$1(a)news.webpartner.dk...
>I am trying to make a system where it would be convenient to convert a
>Windows\DOS filename expression with wild cards to an equivalent Perl
>regular expression.
>
> Examples of DOS filenames:
>
> *.*
> ?er*.sas
> r?p.*s
> e??d*.r*
> ??.?
>
> Does anybody happen to have, or know how to create, a SAS function (or a
> SAS macro) that can do the conversion ?
>
> Anders M�rup
>

comp.soft-sys.sas and www.listserv.uga.edu do not synchronize well, and have
not done so for approx. 14 days.
I estimate that only one third of the mails to the list hits
comp.soft-sys.sas at the moment.

I therefore resubmit the answer from Chang Chung on www.listserv.uga.edu to
my question :

Chang Chung chang_y_chung(a)HOTMAIL.COM wrote:
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>Hi, Anders,This is an interesting question. If you really want to do
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>this well, then it would be *very* difficult. This is because neither
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>windows file system filenames or the perl reg ex are simple. See the
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>gory details yourself at:on
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>filenames:http://msdn.microsoft.com/en-us/library/aa365247(VS.85).aspxon
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>perl reg ex:http://perldoc.perl.org/perlre.htmlIn practice, we can
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>assume away many complications and this will give us a
simpler problem that can be easily solved. Let's assume that your input
filenames have only letters, numbers, a dot(.), and the two wild
characters(*,?), then it is possible to do something like below. Converting
the data step into a proper function using proc fcmp is left as an exercise
for the interested. :-)

Cheers,
Chang

/* assuming that filenames include only letters, numbers, and
a dot(.) as well as the two wild characters. */
data dos;
input dos: $200.;
cards;
*.*
?er*.sas
r?p.*s
e??d*.r*
??.?
;
run;

data prx;
set dos;
length perl $400. d $8.;
len = length(dos);
do i = 1 to len;
c = substr(dos, i, 1);
select;
when (prxmatch("/[a-zA-Z0-9]/",c)) d = c;
when (c=".") d = "\.";
when (c="*") d = ".*";
when (c="?") d = ".";
otherwise do;
putlog "stopped due to an unexpected character, " c;
stop;
end;
end;
perl = catt(perl, d);
end;
keep dos perl;
run;

/* check */
proc print data=prx;
run;
/* on lst
Obs dos perl

1 *.* .*\..*
2 ?er*.sas .er.*\.sas
3 r?p.*s r.p\..*s
4 e??d*.r* e..d.*\.r.*
5 ??.? ..\..
*/
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<Hi
Chang,Thank you for your help.I agree that the perfect solution to the
conversion problem has to be complicated.I would of course have preferred
the "perfect" solution, but your solution seems to be sufficient for my
purpose, if I add a few national characters.Making a SAS function from your
data step solution is a trival task.Best wishes,Anders


 | 
Pages: 1
Prev: HELP: Somers' D
Next: v-fold cross validation in Clustering